Abstract: This article bases on high-precision point cloud data of highways. Firstly, the point cloud data is extracted through the classification processing of point cloud data, and then the point cloud of the tree is projected to the horizontal plane, and the DBSCAN density clustering algorithm is used to realize the extraction of a single tree. Secondly, there is an area where the tree canopy point cloud overlaps in the data-intensive area. This paper extracts the position information of the trunk and calculates all points by combining the geometric features of the trunk The Euclidean distance froms the cloud to the center of the trunk classifies all point clouds to the nearest trunk for coarse segmentation. Finally, the crown model and crown center are determined based on the rough segmented tree contour features. A grid competition algorithm based on density features is proposed to finely segment the overlapping regions. Experiments show that the tree segmentation method used in this paper can achieve accurate extraction of a single tree.

Key words: point cloud data, tree segmentation, DBCSAN clustering, hierarchical grid, mobile measurement system